Är kvinnor sämre på matematik?

"SAT scores even predict success beyond the college years. For more than two decades, Vanderbilt University researchers David Lubinski, Camilla Benbow, and their colleagues have tracked the accomplishments of people who, as part of a youth talent search, scored in the top 1 percent on the SAT by age 13. Remarkably, even within this group of gifted students, higher scorers were not only more likely to earn advanced degrees but also more likely to succeed outside of academia. For example, compared with people who “only” scored in the top 1 percent, those who scored in the top one-tenth of 1 percent—the extremely gifted—were more than twice as likely as adults to have an annual income in the top 5 percent of Americans. länk"

• Betyg (GPA): En metastudie från 2015 finner att korrelationen mellan intelligens och betyg är 0.54 d.v.s. lägre än korrelationen mellan SAT/GRE/ACT och intelligens. Dock är korrelationen mellan en samling objektiva ämnesprov och intelligens 0.81, vilket är lika högt eller högre än testerna ovan.

Hambrick och Chabris, två professorer som jobbar med just skillnader i kognitiva förmågor konstaterar att "SAT correlate very highly with scores on standardized tests of intelligence, and like IQ scores, are stable across time and not easily increased through training, coaching or practice". Något visar att SAT inte på långa vägar är ett test av akademisk kunskap (då hade du kunnat träna upp dig på det). För den som är intresserad av ämnet och vill ha en lättläst introduktion rekommenderas jag deras artikel.

• In whole populations of Scottish 11-year-olds from the 1930s and 1940s there was more variance among males in intelligence by comparison with females, but little difference in mean intelligence; proportionately, more males were at both ends of the intelligence distribution and fewer were in the middle (Deary et al. 2003, Johnson et al. 2008b). This pattern was replicated in a population-representative sample of almost one-third of a million boys and girls tested on the Cognitive Abilities Test in the United Kingdom in the early 2000s: There were trivial differences in mean general intelligence and proportionately more boys than girls at the extremes (Strand et al. 2006). This pattern was replicated in the same test in a population-representative sample of 300,000+ students in grades 3 to 11 in the United States, with the authors concluding that, “The results showed an astonishing consistency in sex differences across countries, grades, cohorts, and test forms” (Lohman & Lakin 2009).

To test this idea, a novel design was applied to data from the U.S. National Longitudinal Survey of Youth 1979 (Deary et al. 2007b). When brother-sister pairs (N = 1,292 pairs) were compared on a general cognitive ability (g) factor derived from the Armed Services Vocational Aptitude Battery and the shorter Armed Forces Qualification Test, the male mean advantage was trivial (less than 7% of a standard deviation), but males once again had substantially greater variance.

Scholars who have assailed the concept of race have forwarded three general arguments against it. Although the arguments are worth consideration, they do not ultimately show that race is a useless or fictional concept. The first two objections are aimed at a straw man, and the last, we will contend, is entirely wrong.

(Objection 1): Human variation is clinal or gradual, not discrete. Skin pigmentation, for example, does not come in four, five, or seven distinct colors, but varies gradually from very dark near the equator to very light in Northern Eurasia.

This charge against the validity of race is undoubtedly correct: a lot of human variation is gradual, not discrete. However, we are not familiar with any prominent proponent of the usefulness of race who would disagree with this contention (assuming they actually understand the evidence). The famous German intellectual and early theoretician of human variation, Johann Friedrich Blumenbach (1775), who is often accused of clumsily categorizing humans into discrete racial groups, contended that, “no variety [of human] exists …so singular as not to be connected to others of the same kind by such an imperceptible transition, that it is very clear they are all related, or only differ from each other in degree.”

For a period of time, polygenism, or the belief that the races arose from separate creations, was popular, but it was widely discredited by genetic and archaeological evidence clearly demonstrating that modern humans originated in Africa (a view promoted by Darwin, who also happened to believe that human races existed). Today, most researchers would agree with Blumenbach, including, for example, Nicholas Wade, who recently wrote a book about race that provoked a furious backlash. In that book, Wade asserted that “because there is no clear dividing line, there are no distinct races — that is the nature of variation within a species. Nonetheless, useful distinctions can be made” (p. 92). This is the key point: although the argument that human variation is continuous rather than discrete is correct, it does not vitiate a sophisticated understanding of race. It only refutes a platonic conception that few contemporary scholars take seriously.


(Objection 2): Human genetic variation is much greater within human populations than among human populations; therefore, variation that exists between groups is of little scientific interest.

This claim is true in a circumscribed sense, but is largely irrelevant to the question of whether population group differences are biologically meaningful. As pointed out by Jeffry B. Mitton and A.W.F. Edwards, the original finding that genetic diversity among human races is insubstantial compared to genetic diversity within races was based on a peculiar way of measuring genetic variation. Roughly speaking, the original claim about genetic diversity was based on analyses at single genetic loci (spots on the chromosome where genes are located) and not on analyses that considered the correlated structure of multiple genetic loci (many locations). Failure to consider multiple loci assures that broad, distinct patterns of allele (gene) frequencies get lost in the noise of diversity at single loci. This sounds painfully abstruse, but the basic point is this: patterns that are nearly invisible for individual genes become visible if one examines multiple genes at the same time (i.e., looks at gene 1 + gene 2…et cetera). <--- därför man med 100% säkerhet kan avgöra vilken ras någon tillhör genetiskt.

Consider a simple but illustrative example.a Imagine that a friend is describing an animal one adjective at a time (e.g., “big,” “furry” et cetera). You are trying to guess the animal. At first, it is difficult to guess because there are many “big” animals, and there are many “big” and “furry” animals. But as her description continues, it gets much easier to guess correctly because each adjective adds to the prior adjectives. The information that allows you to guess correctly does not reside in any one adjective but in the list of adjectives strung together (“big,” “furry,” “antlered,” “white tailed,” “hooved,” “spritely,” “brown,” et cetera). The same holds for population groups. Each genetic locus, like each adjective, is relatively uninformative; but a string of 200 or 300 loci is very informative.

Empirical studies bear this logic out. The geneticist Hua Tang and her colleagues, for instance, found that self-reported ethnicity corresponded almost perfectly with genetic clusters from 326 microsatellite markers (a microsatellite marker is a piece of repetitive DNA in which a series of DNA base pairs are repeated). Other studies have demonstrated even more power to identify people’s ancestry accurately. These studies illustrate that, whatever the meaning of the claim that there is much more variation within than among races, researchers can, if they use the appropriate procedures, distinguish human ancestral groups from each other with remarkable accuracy. The significance of these genetic differences among groups is entirely an empirical question.


(Objection 3): Human racial classifications are arbitrary. For some purposes, categorizing by skin color is useful; for other purposes, categorizing by, say, antimalarial genes, is useful. These classifications, although equally valid, lead to radically different racial categories. Thus, one particular classification scheme is no better than the other and none are particularly illuminating.

By any reasonable understanding of the word “arbitrary,” this claim is incorrect. Perhaps the most prominent proponent of this argument is the gifted and persuasive writer Jared Diamond, who wrote, “There are many different, equally valid procedures for defining races, and those different procedures yield very different classifications” thereby ultimately concluding that we shouldn’t codify human differences into arbitrary racial taxonomies. Diamond is absolutely correct that there isn’t a divinely mandated procedure for correctly classifying human variation. It does not then follow, however, that racial categories are entirely arbitrary, created at the whim of self-interested researchers or racial bigots.

Group categories are constrained by commonly accepted principles such as coherence, parsimony, and predictiveness. Classifications that are incoherent or that have little predictive value are not valid. There will be some flexibility about classification, but not the anarchic freedom Diamond’s arguments seem to suggest. One might, for example, propose classifying Scandinavians with Nilo-Saharan speaking ethnic groups in East Africa because both can digest lactose into adulthood. But, such classifications would violate the principle of parsimony. These groups diverged from each other before developing the ability to digest lactose into adulthood, evolved on separate continents, and do not share other visible traits such as skin pigmentation and hair texture. Therefore, it makes little sense to categorize them in the same ancestral group.

Race, then, is not a platonic essence and racial groups are not discrete categories of humans. Instead, race is a pragmatic construct that picks out real variation in the world (which corresponds to shared ancestry) and allows people and scientists to make useful inferences. In this way, racial categories are like film categories (e.g., drama, horror, comedy). Film categories are certainly real in the sense that they offer predictive power. If one knows that A Nightmare on Elm Street is a horror film, one can be reasonably certain that it will be dark, scary, and violent. But film categories are not immutable essences that perfectly sort movies into distinct types. A genre-based satire like Scream, for example, does not snugly fit into any of the traditional film categories. It might be horror; it might be comedy; it might be some previously unknown combination of the two. Furthermore, there aren’t a fixed number of film categories. The amount and the granularity of film categories depend upon the interests of the people using them. Your friend might use four (horror, comedy, drama, and science fiction), whereas Netflix might use an apparently limitless and startlingly specific supply. (See Daniel Dennett’s book for a variety of points and related examples centering on the topic of species).

The same principles apply to racial categories. If one knows that Thomas is a Caucasian, one can be reasonably sure that Thomas has relatively light skin, and that he has recent ancestry in Europe. But racial categories, like film categories, aren’t immutable essences that perfectly sort humans into distinct groups. There aren’t a fixed number of racial categories, and the number researchers use is partially a matter of convenience. One might start with five continentally based categories (i.e., Caucasians, East Asians, Africans, Native Americans, and Australian Aborigines) and then add more categories as one’s analysis becomes more granular (e.g. Ashkenazi Jewish, Mizrahi Jewish, and so on). These categories aren’t real in some metaphysical sense, but they are useful, and they do have predictive value. In this, they are like many other constructs in the social sciences such as self-esteem, intelligence, and agreeableness. They represent traits that cluster together; they predict outcomes; and they can be quantified.